(33b, c) are not synonymous; (33a) seems neutral between them (Talmy 1978; Gleitman et al. 1996). The difference in
(33b, c)concerns whichofthetwoentitiesis takenas thestandard (or“ground”) against whichtheother (the“figure”)
is measured. Gleitman et al. (1996) argue that this elementof theinterpretation is due to a bias introduced by principle
(31b), not to an inherent asymmetry in the seemingly symmetrical predicates themselves. By contrast, (33a) introduces
no syntactic differentiation betweenChinaandVietnam, so there is nofigure-ground differentiation either.^90
Let me sum up this series of rapid developments. We have come to envision a major consolidation of productive
combinatorialphenomena. We have found thattheformalism ofinterface rulesamong parallelphonological,syntactic,
and semantic structures can be adapted to words, productive and semiproductive morphology, idioms, constructions,
phrase structure rules, and phrasal interface constraints. With luck, then, the onlyproceduralrule will be UNIFY (“clip”
structures together). The counterpart of a traditional“rule of grammar”in this syste mis a lexicalized gra m matical
pattern with one or more variables in it. In order to retain the term“rule”in the traditional sense, I will henceforth
refer to such lexicalized patterns as“1-rules”(“lexicalized rules”) when necessary.
182 ARCHITECTURAL FOUNDATIONS
(^90) A less clear but possible case concerns two related classes of verbs that have been the topic of much discussion over the years (e.g. Postal 1971); Belletti and Rizzi 1988;
Grimshaw 1990; Pesetsky 1994). Note that the pairs in (i) are close to synonymous.(i) a. Jane likes/fears cats.
b. Cats please/frighten Jane.
The assumption is usually that a semantic difference between these verbs predicts their syntactic difference. Yet the difference is relatively subtle. One possibility is that the
conceptual relation expressed by these verbs yields an inconclusive or unstable result in the Linking Hierarchy (Ch. 5, (50)), and that particular verbs stipulate one order or
theother rather arbitrarily. The order chosen thenbiases the interpretationso as to be maximallyin tune withthe Linking Hierarchyand principle (31a). In particular, order
(ia) shades towards a sense of Jane as Recipientof impressions coming to her (i.e.catsas Theme, lower on the hierarchy); whereas order (ib) brings out the sense ofcatsas
Agent, having a psychological effecton Jane as Patient. It is notclear to me whetherthis approachcan accountfor the multitude of curious syntacticeffectsassociatedwith
thisconstruction(seereferencesabove).Butitbearsexaminationas a possibleinstancewhensyntacticargumentpositionisstipulated bya verbinstead ofbeing predictedby
the verb's semantics—and then the syntactic position position biases the semantics as it does in (32) and (33).